The posting
Job Description
The AI Research Scientist will contribute to clinical foundation model workstream for primary-care decision support. The role will focus on adapting open and/or institutionally approved foundation models to Singapore primary-care use cases, such as hypertension and cardiometabolic-renal multimorbidity scenarios involving diabetes, dyslipidaemia, CKD risk, medication burden, preventive care and follow-up planning.
The scientist will prepare structured training and evaluation data from clinical vignettes, synthetic cases, guideline-grounded Q&A, and eHINTS-like longitudinal data extracts where available. The role will support supervised fine-tuning, reinforcement learning or preference alignment experiments, and model evaluation against clinical templates, guideline compliance, safety constraints and usability requirements.
Key Responsibilities
- Prepare model-ready datasets from clinical vignettes, synthetic consultation cases, longitudinal records and guideline-derived tasks.
- Develop and run fine-tuning, instruction-tuning, preference-alignment and/or retrieval-augmented generation experiments for GP-oriented clinical tasks.
- Implement evaluation scripts for guideline adherence, factuality, safety alerts, medication recommendations, missing-data handling and structured output quality.
- Support multimodal and longitudinal data modelling where patient histories, lab trends, medications, notes and care plans need to be represented over time.
- Work with clinical collaborators to translate hypertension and multimorbidity requirements into model tasks, prompts, labels and acceptance criteria.
- Document experiments, datasets, model cards, limitations and reproducibility steps for internal review.
Required Qualifications and Skills
- PhD Degree in computer science, AI, biomedical informatics, computational science, data science or a related quantitative field.
- Hands-on experience with Python and modern deep-learning frameworks such as PyTorch, Hugging Face Transformers, vLLM or equivalent.
- Working knowledge of LLM fine-tuning, prompt engineering, retrieval-augmented generation, evaluation pipelines and reproducible ML experimentation.
- Ability to process structured and unstructured clinical data safely, including tabular records, notes, medications, lab values and guideline documents.
- Good understanding of model evaluation, error analysis, data quality control and responsible AI practices in healthcare or other high-stakes domains.
Preferred Experience
- Experience with clinical NLP, medical LLMs, healthcare datasets, FHIR/EMR-style data, or guideline-based decision support.
- Experience building synthetic data, simulation-based cases, or benchmark datasets for healthcare AI.
- Familiarity with primary-care chronic disease management, especially hypertension, diabetes, lipids, CKD risk and medication safety.
- Experience with GPU-based model training, experiment tracking, and deployment-oriented inference optimization.
